Taking the place-of-worship tagging off the building seems like a bad idea here - the church, not its grounds, is the place of worship. That's how we always tag churches and not doing so means that the building renders in a generic style rather than standing out in the way that you'd usually want a church to on a map.

Princes Street, Cook Street and Marlborough Street are the archetype of what we use highway=pedestrian for. They are pedestrian streets. There are many metres wide, they are flanked by buildings and they are accessible to pedestrians. They literally have "street" in their names. On the ground, they are signed with a white disc inside a red circular border. Many of these roads did, by the way, previously carry vehicles, but that is immaterial to their status as _pedestrian_ streets. Likewise raised kerbs, a red herring. Go to some Italian old towns with narrow streets and see how many of those have separate footpaths. As to highway=footway, this is a valid tagging. It can be used for footpaths that run alongside roads (often above raised kerbs) or through green areas. It can even be appropriate for very narrow alleyways between buildings. You'll see this tagging documented on the wiki, and you'll see that at least one of the sample images resembles the streets I've described: osm.wiki/Tag%3Ahighway%3Dpedestrian

highway=pedestrian is the correct tagging for a roadway with the characteristics of a street and with motor traffic excluded. I am reverting this changeset.

You could be right on this one. The reason I went for a turning circle was that the car appeared to be occupying a widened part at the end of the road, with all points south being more constrained on the west side of the road. But looking more closely, that constraint consists of a short area of bushes (which could overhand anything) and south of that a shadowed region which does actually look like a footpath, meaning no extra width at the end of the road. Since it's such a narrow road, I'm going to remove the turning circle, since the consequences of being wrong are not nice.

On Bing, you can see that the road widens on the left at its end, with a whitish-coloured car occupying the space. It’s widening from a very narrow base, but the purpose of such widening is to assist with turning.

I've remembered where I saw roundabouts with actual refs of their own - it's in Kilkenny on the original ring road: osm.org/way/934089599

I'm not an enormous fan of how the council has applied refs like that to the roundabouts, but it's a clear case of something that should be tagged onto the roundabout itself. Further, such tagging is in conflict with the idea that a transiting road ref could be recorded using the same tag, as now you could be tagging two different kinds of fact with the same tag. This is kind of the dilemma we already have for named bridges that occur along a named road or, to some extent, for named terraces along named roads.

1. It is now. The history shows that it has never had that tag, so clearly the original mapper of the new roundabout missed it and so did I.

2. A rare case, indeed. In this instance, there really is no path around the roundabout where N5 would be potentially wrong at any point of the roundabout. The principle of whether roundabouts should be tagged with the ref of one or more of their participating roads should probably still apply here, so it remains interesting to consider what that principle should be.

3. It's good to look at SatNav as a use case, but I don't buy this specific argument. You most definitely wouldn't want a SatNav in the general case of joining a roundabout to give an instruction like "turn left onto the N5 [context: you hadn't been on it already], drive 15m and then turn left onto the R678. What you would want is "enter the roundabout and take the third exit to the R678". The traveller's awareness of the roundabout in such a context should always be of "the" roundabout, because they have encountered exactly one roundabout and there is no need to distinguish it from others. Further, that instruction that the third exit leads to the R678 should ideally favour not the ref tag of the road that is being joined there, but probably (again, I'm open to discussion on this) any "destination" tagging that is in place, since that ought to reflect the route number and place name(s) provided on the sign the traveller will see when approaching that exit, providing greater reassurance.

On (3), have I overlooked some condition where having a ref tagged on the roundabout itself provides for better routing instructions? It may be better to consider a more general use case where at least 2 routes transit the roundabout on either side of it.

It’s a while since I read the wiki entry for roundabouts and those entries are fallible like the mappers who write them. It has always been my view that a roundabout is a junction and a neutral space. Where they have names, it is a name specific to the roundabout itself. Occasionally they have actual references (can’t spot any just now). In such cases, I would tag these on the roundabout. But in the typical case, if you follow what is on the wiki, you would end up tagging potentially many refs separated by semicolons. To me, this produces a negative outcome. Roads in Ireland don’t multiplex numbers, so you often see interruptions to an N or R route as they join another whose number is used for that section. So to add several refs to a tiny road section is futile. The rendering experience happens to be poor, but of course that can be solved by fixing the rendering. I find it fine to have the roundabouts as members of all relevant route relations (not that we tend to use those). Anyway, that’s why I tag roundabouts as I do, noting that I too am just a fallible mapper. I’m happy to see the topic discussed so we can work out norms that best serve the outcomes we want.

It is not permitted to copy data from other copyright sources. You agreed not to do so when you registered your OSM account. Those Eircodes will have to be removed at considerable inconvenience.

You have infringed the copyright of the Eircode database and, in doing so, undermined the OSM data set. When you registered for OSM, you agreed not to copy data from other maps and data sources, but you have done exactly that. The data you added must now be removed from OSM. This will include any postal codes you may have added, but may perhaps have to include house numbers or street names or any other details that you have infringed from other copyrighted sources. Please identify any other data of this kind that will need to be removed. Similarly, if you have added any data from your own surveying or other personal knowledge, please identify this to avoid its removal. See https://www.eircode.ie/legal (in particular the section on copyright) and https://osmfoundation.org/wiki/Licence/Contributor_Terms#:~:text=This%20means%20that%3A,under%20our%20current%20licence%20terms. for details on why what you have done is not OK. The latter is a set of conditions you explicitly accepted when you registered for OSM.

So Google Maps is another infringement. Don’t copy from other maps! A reminder of the Contributor Terms you ageeed to: https://osmfoundation.org/wiki/Licence/Contributor_Terms#:~:text=This%20means%20that%3A,under%20our%20current%20licence%20terms.

Hi - I see that you have added quite a few Eircodes to the map. Can I ask your source?

In particular, did you source house number information from the same place? If so, they will have to go too.

I was afraid of that. You have infringed the copyright of the Eircode database and jeapordised the integrity of the OSM data set as a result. You cannot simply copy other maps or data sources into OSM and, when you registered, you agreed not to do so. I will remove all of the offending data. Please let me know if you have used other third-party data sources so that I can deal with any other issues. For details of why this wasn't OK, see the copyright section of https://finder.eircode.ie/#/legal

Hi - I see that you have tagged a lot of postal codes on houses. Can I ask your source for these?

That’s random. How did you discover that?

Part of this edit named the footpath "Gutty". Was that intentional?

Hi Daniel,
I could have been clearer with the source in the changeset - I edited the check_date based on Anne being on-site in the last hour and shooting a photo. I've been at this OSM game for rather a long time, so please understand that I'm mindful of the issues you note.

Hi Daniel,
I'm aware of the variety of imagery and the various offsets (back the day, I worked on a system to try to capture local corrections for these) and also of the issues around building "lean". I have been conferring with Anne on the alignment issue for these changes. Your input is useful in terms of the Kilkenny situation, but I'm curious how you find Bing's alignment. In general, I find theirs the most uniformly correct. For this particular set of edits, I chose to leave the alignment as close as possible to where I found them (while separating as required). My intention is to follow up with some local RTK-corrected GPS surveys to capture some known true locations in the area against which imagery layers could be calibrated. In the meantime, the locations of these buildings should be no worse than they were already, and I'm reluctant to invest time in tweaking against an uncertain baseline.

Dermot

Good catch. Fixed.